| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "global_step": 985, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 8.984771573604062e-05, | |
| "loss": 1.1743, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 7.969543147208121e-05, | |
| "loss": 1.0788, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 6.954314720812183e-05, | |
| "loss": 1.0732, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.939086294416244e-05, | |
| "loss": 1.0638, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.9238578680203045e-05, | |
| "loss": 1.0558, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9086294416243655e-05, | |
| "loss": 1.0475, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.8934010152284264e-05, | |
| "loss": 1.0485, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.8781725888324874e-05, | |
| "loss": 1.039, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.629441624365483e-06, | |
| "loss": 1.0402, | |
| "step": 900 | |
| } | |
| ], | |
| "max_steps": 985, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.5601672920445747e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |